Morgan Stanley Interview Guide (2026): Software Engineering & Strats

Morgan Stanley’s campus tech hiring runs through the Technology Analyst Program, and the first gate is a timed coding test: usually two or three HackerRank problems with 60 to 90 minutes on the clock and no interviewer watching. Clear that and you move to a phone screen, then a Superday. Experienced hires skip the assessment and start with a recruiter call and a technical phone screen instead. None of it is exotic. What trips people up is treating a bank like a big-tech shop and prepping for the wrong thing.

Three ways into Morgan Stanley tech

Almost every technology candidate lands in one of three buckets, and the interview leans differently depending on which one you’re in.

The Technology Analyst Program is the new-grad pipeline. You apply to a broad tech role, not a specific team, and get matched after you’re hired. Analysts land in application development, infrastructure, cybersecurity, or platform teams across the firm’s tech hubs: New York, Montreal, Glasgow, Budapest, Mumbai, Bengaluru, and Shanghai. Montreal is worth knowing about specifically — the firm opened it as a dedicated technology center and has since grown it past 2,700 people working on cloud infrastructure, cybersecurity, and machine-learning tooling for every division. Glasgow is older still, dating back to a settlements team of six the firm started there in 2000, and it’s now one of the larger engineering sites outside New York. Some analysts land on wealth-management engineering, which absorbed a chunk of retail brokerage technology when Morgan Stanley bought E*TRADE in 2020, or on asset-management platforms like Parametric, the custom-portfolio unit whose engineering org runs out of Alpharetta, Georgia. The bar here is standard data structures and algorithms plus a behavioral read on whether you actually want to work at a bank.

Strats and Modeling is Morgan Stanley’s version of the quant-adjacent developer. Strats sit between the trading desk and the codebase: they build pricing models, risk engines, and P&L tooling, and they’re expected to reason about the math, not just implement someone else’s spec. If you’re interviewing for strats, expect probability and expected-value questions mixed in with the coding. Python and C++ dominate here, with some kdb+/q where tick data lives.

E-trading technology is the low-latency world: order management systems, matching logic, market-data handling, execution algos. This is the group behind the firm’s electronic trading infrastructure, internally referred to as MSET, and its ultra-low-latency co-located execution platform. The questions get more systems-flavored and more concurrency-heavy. C++ still rules here, and a question about an order book is a genuine test, not a warm-up.

The HackerRank round comes before any human reads your code

The HackerRank round is auto-graded on hidden test cases. Nobody reads your code for style. That means two things: your solution has to actually pass the edge cases (empty input, single element, duplicates, large N that times out an O(n²) approach), and partial credit is real, so a brute-force answer that passes half the cases beats a clever one that compiles wrong. Budget your time. If a problem is fighting you at the 20-minute mark, lock in a working brute force and move on.

The problems themselves are LeetCode easy-to-medium. Arrays and hash maps, two pointers, a string-manipulation problem, sometimes a light graph or dynamic-programming question. You won’t get a research-grade hard problem in the automated round.

Phone screens and the Superday

The phone screen is 45 to 60 minutes, usually a working engineer on the other end, and it’s a live coding problem shared over a collaborative editor. Talk through your approach before you type. They care about how you narrow the problem, name your assumptions, and reason about complexity. A silent candidate who arrives at the right answer often scores below one who’s clearly thinking out loud and gets 90% of the way.

The Superday is the final round: usually four or five interviews packed into one day, mixing technical and behavioral, sometimes with a hiring manager and a more senior engineer or executive director in the loop. For 2026 these are often virtual for early rounds with an on-site final, though it varies by office and team. Expect at least one interview that’s mostly “why Morgan Stanley, why finance, tell me about a time you disagreed with a teammate.” Banks screen hard for people who’ll bail after the training program, so a credible answer to “why here and not a startup” carries weight.

The coding problems that recur most on the software track

The coding is bread-and-butter DSA. A representative spread for a software track:

• “Given an array of daily prices, find the maximum profit from a single buy and later sell.” This is the classic best-time-to-buy-and-sell-stock: sweep the array once, tracking the lowest price seen so far and the best profit you could book against it, for O(n) time and O(1) space. The trap is a double loop that times out, or forgetting that a sell has to come after its buy.
• “Return the first non-repeating character in a string, and give me the time and space complexity.” Count every character with a hash map in one pass, then scan the string again and return the first with a count of one — O(n) time, O(k) space for the alphabet. Interviewers watch whether you assume ASCII or handle a larger character set, and whether you actually state that space cost instead of hand-waving it.
• “Reverse a singly linked list in place. Walk me through the pointer bookkeeping.” The clean approach is three pointers — previous, current, and next — flipping one link per step so you never lose the rest of the list. Expect a follow-up asking for the recursive version or for reversing only a section between two positions.
• “Detect whether a string of brackets is balanced.” A stack is the entire solution: push each opener, pop and check the match on each closer, and accept only if the stack ends empty. The cases they probe are a wrong-type match like “(]” and a closer that arrives with nothing on the stack.

For experienced software and e-trading candidates, at least one round goes design-heavy. These are asked conversationally, not as a spec: “Design an in-memory limit order book. What structures give you constant-time best bid and ask plus fast cancels?” or “A market-data feed pushes millions of messages a second and a downstream consumer falls behind. What do you do?” They want to hear about ring buffers, back-pressure, dropping stale ticks, and where the latency actually goes, not a textbook recitation of load balancers.

The probability bar on the Strats loop

If you’re on the strats track, the probability questions are the differentiator, and they’re closer to a quant screen than a normal SWE loop. Nothing here needs measure theory, but you need to be fast and clean with expected value and conditioning:

• “You draw cards one at a time from a shuffled standard deck, without replacement, until you see the first ace. What’s the expected number of cards you’ll draw?” (It’s 53/5, or 10.6.) The clean route in: the four aces split the deck’s other 48 cards into five gaps of equal expected size, so on average 9.6 non-aces show up before the first ace; add the ace itself and you get 53/5. Interviewers want that symmetry argument, not a brute-force sum over 52 terms.
• “Two players alternate flipping a fair coin; first to flip heads wins. What’s the probability the first player wins?” (2/3.) Set it up as one equation: the first player wins immediately with probability 1/2, or both miss and the position resets, so p = 1/2 + (1/4)p.
• “A stick is broken at two uniformly random points. What’s the probability the three pieces form a triangle?” (1/4.) The pieces form a triangle only when no single piece is longer than half the stick; plot the two break points on a unit square and that condition covers a quarter of the area.

The pattern: set up the states or the geometry, write the equation, solve. If you can narrate that cleanly you’ll pass even if you fumble the arithmetic. Interviewers on strats desks have seen every canned answer, so showing the derivation matters more than sprinting to the number.

Morgan Stanley’s engineering culture: stack, rank, and hours

It’s a bank, and that shows up in the day-to-day. The stack skews Java across the firm, with C++ where latency matters and Python heavy in strats and data work. Some codebases are old and load-bearing; some teams have modernized onto cloud and modern tooling. Internal platforms are strong because the firm has been building them for decades, which is a double edge: powerful, well-documented systems you couldn’t get anywhere else, wrapped in process you also couldn’t get anywhere else.

The hierarchy is explicit. Analyst, associate, vice president, executive director, managing director, and the ladder moves on a schedule more than on a rocket. Promotions are title-gated and slower than the “senior in three years” pace common at big tech. As of 2026 the firm is largely back in the office most of the week, so if remote work is non-negotiable for you, weigh that going in. Work-life balance in technology is generally better than in banking or the trading floor, but engineers on production trading systems carry on-call, and a market-hours outage is a real fire.

The upside is stability, genuinely interesting problems in trading and risk, and colleagues who know the domain cold. The trade-off is that you’re one function inside a large financial firm, not the product itself. If you want to ship a consumer feature and watch a graph move, this isn’t that. If you want to build the system that prices a book of exotic derivatives, it very much is.

Comp: lower base than big tech, and the bonus carries the story

Bank technology pay trails big-tech total comp, and the gap widens as you climb because big-tech packages lean on stock that compounds while bank pay leans on cash bonus that resets each year. The base salaries are competitive; it’s the equity that’s thinner. Bonuses are a larger and more variable slice than most tech roles, and at senior levels a meaningful chunk converts to restricted stock that pays out gradually over the following few years rather than landing as cash all at once.

Treat the numbers below as rough US ranges for 2026, not quotes. They swing hard by location, team, and how the firm’s year went. Check levels.fyi and recent Glassdoor entries for the specific office and level before you anchor on any figure or counter an offer.

Level / role (Morgan Stanley technology)	Approx. US base salary, 2026	Typical annual bonus	Bonus form
New-grad Technology Analyst	$95k–$130k	10%–25% of base	Almost all cash
Associate (~3–6 years)	$130k–$180k	15%–40% of base	Mostly cash, small deferred piece
Vice President	$180k–$260k	25%–70% of base	Cash plus deferred stock
Strat / quant developer	Overlaps associate–VP bands, often higher variable	Wider swing, tied to desk performance	More deferred at senior levels

For a rough comparison: a new-grad software engineer at a top big-tech firm often clears $150k–$200k+ all-in on day one thanks to a large stock grant, while a Morgan Stanley tech analyst lands closer to $110k–$160k with most of the variable piece in cash. The bank number is more predictable and less exposed to a stock price; the big-tech number has more upside and more risk. Which you prefer is a real choice, not a mistake either way.

Where to spend your prep time before the loop

Get fluent on easy-to-medium LeetCode across arrays, strings, hash maps, two pointers, linked lists, stacks, trees, and basic graph traversal. You don’t need the hard-problem grind that a big-tech loop demands; you need speed and correctness on the common patterns, since the automated round rewards passing tests over elegance. Pick one language and know it cold, including its standard collections and their complexity. For most software roles that’s Java; for e-trading, C++; for strats, Python.

If you’re going for strats, spend real time on probability. Work through expected value, conditioning, coin and dice problems, and a handful of classic brainteasers until the setup is automatic. A book like Heard on the Street or A Practical Guide to Quantitative Finance Interviews covers the exact flavor.

Then do the unglamorous part: read up on the firm’s actual businesses (wealth management, institutional securities, investment management) so your “why here” answer isn’t generic, and prepare two or three concrete stories about projects where you owned something hard. The technical bar at Morgan Stanley is passable for a prepared candidate. The people who fail the loop usually aren’t the ones who couldn’t reverse a linked list. They’re the ones who couldn’t say, convincingly, why they wanted to build software at a bank rather than anywhere else.